Berlin workflow automation platform with a visual editor, native AI agent nodes and Model Context Protocol support, also for HR.

UI-TARS Desktop overview and key facts (October 2026)
Open-source agent stack from ByteDance that controls computers and browsers with a vision-language model via mouse and keyboard.
What is UI-TARS Desktop?
UI-TARS Desktop and Agent TARS are open-source agent projects from ByteDance that let a multimodal model operate a computer or browser by looking at the screen and using mouse and keyboard, for automating tasks in applications without an API.
The repository contains two parts. UI-TARS Desktop is a native app for Windows and macOS that uses the UI-TARS vision-language model for precise mouse and keyboard control. Agent TARS is a general multimodal agent stack with a command line and web UI that controls browsers through visual grounding or the page structure and connects tools through MCP servers. It supports several model providers, including Anthropic and Volcengine. The GitHub API showed about 39,000 stars and a push on September 24, 2026.
UI-TARS Desktop AI features and use cases
- Automates software without an APIThe agent reads the screen and clicks, types and scrolls like a person. Teams can automate steps in older desktop applications or web portals that offer no interface for integration.
- Runs browser tasks from a text instructionAgent TARS controls a browser through visual grounding or the page structure. Research, form filling or data collection across websites can be described in one sentence and then reviewed.
- Extends agents with MCP toolsIts MCP-based architecture connects additional tools. Developers can combine screen control with direct access to files or services.
What sets UI-TARS Desktop apart
- Pairs an agent app with ByteDance's own UI-TARS vision-language model for GUI control.
- Offers both a desktop app for Windows and macOS and a command line and web UI stack.
Where UI-TARS Desktop is strongest
Open-source software with no country restriction that runs on your own infrastructure anywhere.
Questions to ask before you buy UI-TARS Desktop
An agent that controls mouse and keyboard can do anything the logged-in user can, so test it on isolated machines and with restricted accounts first. Screenshots may contain personal data and are sent to the chosen model provider unless you host the model yourself. Check the model license and the terms of each provider.
UI-TARS Desktop reviews
No reviews yet. Be the first to share how your team uses it.
Every review shows its verification level. We read each one before it goes live. How we check reviews
Questions about UI-TARS Desktop
Which operating systems does UI-TARS Desktop support?
According to the README, Windows and macOS, plus browser operation.
Which models can it use?
It works with the UI-TARS model and supports several providers, including Anthropic and Volcengine, according to the README.
Is it free for business use?
The code is released under the Apache 2.0 license. Model usage may cost money depending on the provider.
UI-TARS Desktop update historyno updates yet
No changes recorded for UI-TARS Desktop yet.
Changes the vendor made and confirmed user reports, each checked against the vendor's own pages. We review every tool monthly.
Alternatives to UI-TARS Desktop
Tools in the same category (Agents & automation) with the most overlapping markets.
Open-source framework that lets language models operate a web browser on their own, as a Python library, CLI or hosted cloud.
Open-source zero-code platform from OpenBMB in which several AI agents work together on software and other tasks.
Open-source visual builder for AI agents and RAG flows that can be deployed as an API or MCP server.



